Thank you for all of the detailed help and great software and apologies for such basic questions, this is all very new to me.
I want to conduct a meta-analysis of functional data on a small group of studies as not much research has been done in the area.
These activations will then be compared to activity for perhaps two other tasks but these will have much more research available. (Will try and make these tasks as specific as possible to attempt to equate the number of studies used in each group if this is the best approach?)
I am currently identifying possible studies then believe I can include them if:
- they report in MNI or talairach space
- they have whole brain analysis
Are these the only necessary requirements? 
Am I right in thinking that it is better to use the group-level statistical parametric maps as I have a small number of papers?
I assume that I need to contact the original authors to obtain this information, is that correct?
If so what information exactly do I need to ask for?
I'm aware there is also an issue of the same statistical threshold being required across the contrasts reported in the paper. Is this still relevant if using the statistical parametric maps?
Any additional information on the practical steps needed to get all the information required to begin the analysis would be much appreciated.
Apologies again if all of this should be very obvious!

Dear ___, please see responses below:

> Will try and make these tasks as specific as possible to attempt to equate the number of studies used in each group if this is the best approach?
Well, if you are comparing only two groups, the more studies the more statistical power...

> Are these the only necessary requirements?
An other important criterion is that the statistical threshold is the same for any voxel !

> Am I right in thinking that it is better to use the group-level statistical parametric maps as I have a small number of papers?
Maps are always better.

> I assume that I need to contact the original authors to obtain this information, is that correct?
Yes...

> If so what information exactly do I need to ask for?
In SPM, these maps are usually named ...spmT... In FSL they are usually named ...tstat...

> I'm aware there is also an issue of the same statistical threshold being required across the contrasts reported in the paper. Is this still relevant if using the statistical parametric maps?
No, because these maps are not thresholded. Note that the criterion is that the statistical threshold is the same for any voxel.

Dear Joaquim,

Thank you for your previous responses. 
I've read in published work and elsewhere on this forum that there may be issues including multiple contrasts from the same participants but am unsure whether this applies to my situation.
I will be comparing two groups of studies formed based on distinct tasks. However 3 of the studies (I have just over 20 in each group) have a task in each category so could qualify for both.
Do I sacrifice some power in the analysis by including them in both groups?
I am not desperate to do so as I have over 20 studies and only including them in the group my main focus is on means a more unbalanced number of studies between the groups but would appreciate your advice.

Many thanks,
Jo

Dear Jo,

I would suggest to include the studies in only one group.

Best wishes,

Joaquim